I’ve already had an extra cup of coffee this morning, so I’m ready to walk through this.
https://www.stereophile.com/content/hot-stuff-loudspeaker-voice-coil-temperatures-page-2. (The first page gives some back ground, but you can skip the first page, and go directly to page 2 for the details of the test.)
Everyone agrees that a speaker voice coil heats up when used. The louder the playback, the more it heats up. It is also known that a speaker voice coil, when hot enough, conducts electricity less well than if cool. This is said to result in a loss of output, especially in tweeters, and is called thermal compression or power compression. Apparently this is a known problem in auditorium PA loud speakers. Many people also believe that thermal compression occurs in home audio speakers.
The question is, do home audio speakers suffer the same loss of output to voice coil heating? Does the resistance of woofer or tweeter voice coils increase with use?
Keith Howard, the author of the Stereophile article, measured the voice coil resistance in a B&W CDM1NT speaker that he owns. He used them as an example of compact 2-way speakers of lower than average sensitivity, in which you might expect to find thermal compression problems. He also had physically removed the speaker’s crossover board from the cabinet, allowing him to measure thermal behavior downstream of the crossover.
He had to build some test equipment so he could directly measure the voice coil temperature (did he mean resistance?) of both the mid-woofer and the tweeter.
As program material, he wanted a music passage that was likely to heat the speaker’s voice coils. He quoted a good authority (Bob Stuart of Meridian) that Metallica tracks make for good power-testing material, as they have negligible dynamic range and quite a wide frequency spectrum. As he didn’t have any Metallica, instead he chose a passage from the song Heartbreaker from Led Zeppelin II. He claimed, the instrumental section in the latter half of that song was just what he was looking for, with a paucity of dynamic range, and quite a wide acoustic spectrum (fig.2). Because this passage lasts about 47 seconds, he stitched seven repeats (329 seconds) end to end in a WAV file to create a single track of over five minutes long.
This graph (figure 2) shows two traces of dB full scale (dBFS) vs frequency of peak (red) and average (blue) audio spectra of the 47 second passage. The author claims the wide frequency range and close spacing of the red and blue traces confirm the low dynamic range of the passage. I’m glad he at least tried to document the dynamic range. Without anything to compare to this, we have to take his word. If anyone cares enough, they can do a similar measurement with another musical passage.
Playback was done with a Townshend TA565 universal disc player and a Rotel RA-1062 integrated amplifier rated at 60 Wpc continuous into 8 ohms. The volume control was set to about 2 o'clock, at which the sound-pressure level at 1 meter driving the CDM1NT's mid woofer driver read between 100 and 104 dB. The author described the mid-woofer’s cone as “undergoing sufficiently large excursions to threaten the end-stops had the volume been much higher”. He also measured the tweeter at the same volume setting.
If the author directly measured voice coil temperature, he didn’t show those results. Instead, all his results show voice coil resistance in ohms over a time of 500 seconds, where the seven back-to-back repeats of the music took place during the first 329 seconds.
The results (figures 3 and 4) were something of a surprise. Despite what he thought was a high playback level, the increases in voice coil resistance were even less than he expected in the mid-woofer and barely detectable in the tweeter.
Figure 3 – mid-woofer
The trace began at 3.9 ohms before the music passage began, rising to about 4.2 ohms while playing the test program, and rapidly returned to 3.9 ohms afterward.
Figure 4 – tweeter
Same test as in figure 3, but with the tweeter. Note that the vertical scale is different than in figure 3. It shows the range of 2.5 to 3 ohms, over the same vertical distance as the 3.5 to 5 ohms in figure 3. It is ‘zoomed in’ or magnified. Here the change in resistance during the test program was barely distinguishable, amounting to about 0.01 ohm.
Conclusion: Under these test conditions, thermal compression is barely detectable, and is not considered a problem.
